Nofollow and Wikis

    January 12, 2008

There is no such thing as a free link.  No matter what the person who is selling you it tells you.  And the web will always adapt to make it so.  Even if its hard work.

Barry Schwartz posted Get A Free Link From Wired today on an SEO blog, noting that:

Some SEOs were saddened when Wikipedia added nofollows to external links. Perhaps they’ll perk up to discover that Wired’s semi-Wikipedia challenger has no such blocking

The post proceeded to give advice and instructions for how to spam the Wired How To Wiki.  And it has been by the SEO community.  As a member of the community and working at the provider of its most excellent wiki software, I edited out some of the spam as it didn’t meet the goals of the community.  But its not that simple.

I believe the intent was to point out an exploit, and not necessarily for bad.  The problem rests with nofollow being a good tool for some purposes (blog comment spam) and not for others (wikis in general).  I had a quick email exchange with Danny Sullivan and he also talked with Wired and edited the post:

NOTE FROM DANNY: We’ve talked with Wired about the situation, and they are putting a robots.txt block on links coming out of the wiki so that links won’t pass credit. Also, our apologies to Wired in that we’ve ended up causing a run on the wiki with new pages being created. That was definitely not our intent — the headline of getting a free link, and the article itself, was more tongue-in-cheek about how the system was and might further get abused, rather than advice for people to really misuse the wiki for promotional purposes. I don’t agree with that type of abuse in general, and as someone who has had to deal with it in comments or submissions to our forums, it’s no fun. In hindsight, we probably should have just dropped a note pointing out the vulnerability. We’ve also asked that our test page be completely removed — it has served its purpose now.

If every website was a wiki I might have edited something similar on their post while disabling the link.  Later over a drink with Paolo Valdermin I was joking that we should invent the unfollow link, or make it so any traffic coming from the site doing damage was rejected.  But I’m joking and need to digress into a topic most people don’t understand:

unfollow was a byproduct of Vote Links, created by Kevin Marks, which I believe I had a hand in at least inspiring.  It was designed as a tool for making blog comments not count in Google’s PageRank, while letting the blog post’s links count.  Blog comments are relatively good at dealing with comment spam because of what Clay Shirky described as encapsulation — every blog has an owner who can determine how to moderate their comments.  nofollow is just one way. 

Now comments with nofollow enabled get spam anyway, maybe because the fact that the blog host is communicating behind the scenes with search engines is ignored, or that SEO isn’t the only goal of the spammer.  If you own a blog, you deal with manually sorting through vandalism all the time. 

Still, nofollow is a good thing for some Social Software.  But it doesn’t belong or work everywhere.  Can you imagine a web that works as good as today’s if every link was tagged with nofollow?

Wikipedia enabled nofollow, much to the chagrin of the SEO community, for some good reasons that also need additional context:

  • Wikipedia is an exception to any other wiki community on the web
  • Wikipedia has no feature to detect and delete spam, instead it is a feature made of people, who are better at such decisions when one of the goals of the system is openness
  • Wikipedia’s core editors are extremely burdened with this task because of how valuable its attention is
  • The result is Wikipedia gets lots of link love, but doesn’t give any to anyone

IMHO, Wikipedia made a mistake implementing nofollow.  While blogs have one or a couple authors, a few commenters, and many readers — wikis have many authors and many readers.  In an a wide open wiki, just one possible configuration, you can’t tell who is trustable or accountable given the missing identity and reputation layers of the web.  nofollow doesn’t work for wikis because:

  • Wikis are part of the web of links
  • I haven’t seen a partial and prudent implementation of nofollow in a wiki.  And its hard to picture one that doesn’t discriminate upon users automatically.  You could make nofollow apply to links from "trusted users" but different communities should develop different rules for trust.  And such rules could kill communities, especially nascent ones by hard coding rules too early.
  • There are other wiki spam countermeasures
  • If you enable nofollow, every link gives no value, regardless of its value, and that value is best defined by the community using it

Part of how wikis deal with spam is the phantom authority — there is a higher transaction cost for damaging a wiki than fixing it.  And people are generally good.

That said, even though I may not want to be in business with a web like this, I care about what’s been created on this wiki.  And so I’ll be spending some of my time this weekend reverting vandalism.  Better that than giving in by shutting it down, or uninformed suggestions to bluntly implement nofollow.  Besides, the irony is that some of the SEO posts actually have some good How To content that actually needs a little human editing.

Some links to SEO Posts about all this (maybe they will pitch in if I link to them): WebProNews, Wired News and Mashable!